At this point, it’s hard to tell if former president Donald Trump’s conspiracy theories about the 2020 election are old or new. Everything is sort of blurry, both because so many of his claims have been completely debunked, prompting him to occasionally offer them more vaguely, and in part because Trump always does this with rhetoric. He takes some point he thinks is great and, through repetition, wears it down to a catchphrase that close observers immediately understand. (Do you know who the “FBI lovers” are? Congratulations, you’ve seen this effect at work.) Many of his unfounded claims about fraud in the 2020 election simply require one touchstone — “Arizona!” — and everyone gets the drift.

But then there are more specific things such as this line of attack he offered during a rally in Iowa over the weekend.

“Joe Biden lost with the African American population to former president Barack Hussein Obama in every single state other than the five swing states, in other words, the only states that were going to determine the election — although I think a lot of other states were corrupt also.”

This particular allegation is accusatory on two levels, of course. There’s the actual allegation, that President Biden underperformed with Black voters everywhere but in swing states, which we’ll get to in a moment. Then there’s the subtext, if you can use that term to describe something glaringly obvious. Oh, so there was fraud done by Black voters who’d voted differently for Barack Hussein Obama, you say? I certainly wonder how that will be interpreted by Trump supporters?

It’s not new that Trump would make allegations of fraud that impugn Black people in particular. When he first started making broad allegations of fraud, as he trailed Hillary Clinton late in the 2016 contest, he focused on voters in Pennsylvania, by which he obviously meant voters in Philadelphia, by which he obviously meant voters in heavily Black parts of the city, as a popular conspiracy theory about 2012 had it. At this late date, we probably don’t need to spend a great deal of time adjudicating Trump’s history with allegations about race, so we won’t.

But it’s useful to assess this claim made in Iowa simply because it’s 1) interesting and 2) misleading. It’s just this lazy sort of “Black Americans were up to something” that should be considered in the context of having come from the leading political voice in the Republican Party.

What Trump is alleging is that Obama outperformed Biden with Black voters everywhere but in five swing states, presumably referring to those that flipped against Trump in 2020: Arizona, Georgia, Michigan, Pennsylvania and Wisconsin. It’s not clear what exactly this means or what data set Trump is ostensibly using — Biden did worse than Obama in 2008? 2012? According to exit polls? Some proprietary voter data? — but we can nonetheless try to assess it.

Exit polls do show that the level of support for Biden among Black voters in four of those five states (there were too few Black voters in Arizona to measure) was higher than the national level of support he saw from Black voters. But the question is whether he outperformed Obama in those states with Black voters and, according to exit polls, he didn’t. In both 2008 and 2012, Obama got more Black support than did Biden in each of those four states.

2020
2008
2012
National
87%
95 (+8 over 2020)
93 (+6)
Arizona
no data
97
no data
Georgia
88
98 (+10)
no data
Michigan
92
97 (+5)
95 (+3)
Pennsylvania
92
95 (+3)
93 (+1)
Wisconsin
92
91 (+1)
94 (+2)

Regardless, exit poll data on demographic subgroups such as this should be taken with a healthy dose of salt. For example, exit polls suggested that Biden got only 87 percent of the Black vote; polling from Pew Research Center validated against voter data indicated that Biden got more than 90 percent support.

So let’s instead look at correlations between the density of the Black population in counties and how they voted in 2020. If we break out counties into 10 groups, clustered by the density of the Black population, we can see how swing states compare with other states in how their votes changed since 2008 and 2012. It looks like this; rest assured, an explanation follows.

The question is where the orange line (swing states) sits relative to the magenta line (other states). And toward the left side of each graph (counties where the density of the Black population is lower) Biden fares worse. In more heavily Black counties, though, Biden often gained more votes relative to Obama in non-swing states (magenta line higher than the orange one) — with the exception of the most densely Black counties. In all counties, there was an average shift in the vote margin to Trump, but that shift was smaller in more heavily Black counties and slightly smaller in swing counties that were more heavily Black.

Complicated! So let’s just break out the 242 counties that are at least a third Black. There, the number of votes cast for Biden was, in fact, higher in swing states than in non-swing states relative to votes for Obama in 2008 and 2012. The shift in the vote margin was about even.

Aha! Trump vindicated! Well, no. First of all, Biden did better than Obama with Black voters in non-swing states, which isn’t what Trump had asserted. Second, if we simply add up the vote change in the counties, the picture offered above changes. Heavily Black counties in swing states cast about 223,000 more votes for Biden in 2020 than for Obama in 2008 — but heavily Black counties in non-swing states cast about 235,000 more votes for Biden. The difference was in the votes cast for the Republican. In swing states, those counties cast 95,000 more votes for Trump than Obama’s 2008 opponent, John McCain. In non-swing states, they cast about 14,000 votes fewer.

The ratio compared to 2012 was different. Biden got about 326,000 more votes in 2020 than Obama did in 2012 in heavily Black counties in swing states, compared with an improvement of 185,000 in non-swing states. But, again, Trump improved over Mitt Romney by more in swing-state counties, earning 128,000 more votes in heavily Black counties in swing states.

A big part of the shift in 2012 (and 2008, for that matter) is a function of Fulton County, Ga. — home to Atlanta. A lot more people voted in Fulton County in 2020 than in 2008 or 2012, most of them for Biden. Of course, there are also a lot of people in Fulton County, meaning that a shift in support can have a big effect. As a percentage of votes added in each county on average, Biden saw more improvement in heavily Black counties in non-swing states.

If we break out the counties by type (using Pew Research Center’s definitions), you can see better how the numbers are skewed by big shifts in larger counties.

But you can also see how there are big shifts in a lot of non-swing states, too.

There are of course other considerations. The population of the United States grew by 9 percent from 2008 to 2020 and last year saw a huge surge in interest, generally predicated on opposition to or support for Trump. That Obama was the first Black president was, of course, another relevant factor.

All of this is admittedly very much in the weeds on a proclamation from Trump that there’s no reason to consider in good faith. It’s also obviously false; the only candidate in 2020 who earned fewer votes than his party’s candidate in 2008 or 2012 was Trump, who was outperformed in heavily Black counties in swing states by McCain (though Biden still beat Obama’s totals in those counties). It’s an interesting thought experiment, in other words, but one that distracts from the immediate impression Trump intended to leave with his claim: Black voters were up to something.

That message can be evaluated on its own merits.